.png){kind=icon}
.png){kind=icon}
Return
Defect Orientation Insensitive Depth Estimation Method Using a Novel Differential Eddy-Current Probe
.png){kind=icon}
PRE
AI
DOI：10.1109/jsen.2026.3706800.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Eddy-current testing (ECT), as an efficient noncontact nondestructive testing (NDT) technology, has been widely applied to defect detection in critical components across aerospace, nuclear power, and additive manufacturing industries. In this article, to address the challenges in manual blind testing where both the defect depth and orientation would affect the signal response during detection, a novel defect depth estimation approach is proposed with an orientation-insensitive sensor probe and machine learning method. A differential eddy-current sensor based on rectangular excitation coils and L-shaped pickup coils is designed. Its key parameters are optimized through finite element simulation to obtain the optimal structure for testing. Feature selection is performed on the scanned voltage signals using the random forest (RF) model, and defect depth reconstruction is achieved using the CatBoost classifier. Experiments have been conducted, and the results demonstrate that the proposed method achieves an average accuracy of 95.42% in defect depth inversion.
Keywords:
CatBoost classifier
defect depth estimation
differential sensor
eddy current
L-shaped coil
nonintrusive
Journal
IF:
4.5
Papers:
2.1W
Citations:
7.3W